Kouper Health

Conversational AI drives inbound requests and outbound patient outreach across the Primus, Archie and Cura agents. The platform also maintains shared patient context and coordination workflows, providing value beyond automated conversations.

Teams agree before launch on the tasks in scope and who receives escalations through their existing EHR. Agent actions are logged. Numerical escalation thresholds, overrides and recovery procedures after an incorrect action remain unspecified.

Generative AI supports voice and text interactions using shared patient context across workflows. The available architecture does not identify production models, suppliers, training datasets, model versions or how changes are evaluated before release.

The privacy policy identifies categories including AI, messaging, hosting and analytics providers without naming them. It does not map patient data to particular model processors or specify those processors' retention and training terms.

Named partners include Lifepoint, Mosaic and Ascension Saint Thomas. Reported gains exceed 50% for engagement and 60% for follow-up appointments, but lack patient denominators, measurement periods and a comparison method linking results to a specific deployment.

Patient conversations can contain protected information under partner agreements. Model training use, retention windows and patient data deidentification rules remain unspecified. The policy's general development section concerns partner representatives and website visitors; it does not settle patient data treatment.

Kouper acts as a HIPAA business associate for patient communications. Its service privacy policy covers diagnoses, care requirements, scheduling and message content. The BAA itself, contract tier and subcontractor provisions are not available to inspect.

The current security FAQ describes access controls, secure handling and auditability. It does not establish an independently checked security program, a current certification scope or a retrievable vendor audit artifact.

The text service arranges care and follow-up rather than providing medical advice or treatment. Licensed providers retain the clinical relationship. Risk tracking supports navigation, without establishing a cleared diagnostic or treatment function.

A safety framework and action logging support oversight, while deployment planning defines ownership and call scope. A formal evaluation protocol, subgroup performance and independent audit of model behavior remain unspecified.

Action logs make agent activity traceable for care teams. Consumer text service terms disclaim warranties and limit liability. An output correction obligation, accuracy guarantee or compensation route tied to agent mistakes is unspecified.

The integration FAQ describes native connections to hospital and ambulatory EHRs, and escalation tasks can enter existing staff workflows. Named systems, supported standards, read and write boundaries and implementation details remain unspecified.

Patients interact through calls and text messages connected to provider workflows. Hosting providers, deployment options, data storage regions, tenant isolation and cross border processing are unspecified, leaving the infrastructure arrangement unclear.

Commercial access begins with a tailored demonstration. Subscription rates, charging units, minimum commitments and implementation amounts are unspecified. Patient messaging carrier charges in the consumer terms do not establish the organization's platform cost.

Named hospital partners support transitions from inpatient and emergency care into follow-up. The broader specialty range includes primary care, oncology, pediatrics and dentistry, without equivalent validation for every specialty and agent workflow.